Abstracts

Rationale: Pre-surgical work up for resective surgery or neuromodulation device insertion in some patients with drug-resistant epilepsy involves invasive electrographic recording to better delineate the seizure onset zone and/or to map eloquent cortices. Understanding the exact location of each electrode contact in relation to cortical and sub-cortical landmarks is crucial during the interpretation of the recording. Currently, many clinicians use a manual labeling approach, and some algorithms are available for grids and strips; however, there are no reliable automatic options for the stereotactic EEG (SEEG).

Methods: The only framework we found that could robustly tackle this problem and account for the inherent noise is RANSAC, an iterative method to estimate a mathematical model’s parameters from a set of observed data that contains outliers. However, due to the inherent randomness in the algorithm, existing electrodes can be missed. To tackle this problem, we implemented a novel algorithm to detect the electrodes’ penetration points from the post-op CT. We use the marching cubes algorithm, a high-resolution 3D surface construction algorithm, to detect the skull (CT-skull), which has a very high intensity in the CT scan. We then cleaned the marching cubes algorithm’s noisy output by keeping only the connected surface with the largest area and simplified the surface using the quadric error metrics algorithm. To detect the penetration points, we overlayed the skull reconstruction from the pre-op MRI scan. The MRI-skull surface was segmented from pre-op MPRAGE images using the watershed algorithm. Then, the post-op CT was registered to the pre-op MRI using a boundary-based algorithm. To find the sEEG penetration points, we took the parts from the CT-skull that pop out from the MRI-skull. We automatically grouped these vertices using an algorithm that clusters connected triangles on the surface. The last step was to calculate the centroids for each cluster, which represents the sEEG penetration points

Results: Without a priori number of electrodes and contacts, the algorithm detected all the sEEG penetration points in 32 patients (13±3 electrodes and 156±52 contacts per patient). The hours-long task of manual identification was reduced to minutes per patient.

Conclusions: We present a new method for automatic detection of SEEG electrodes’ penetration points. We show a 100% success rate in finding the true penetration points. In the future, we are going to combine this information with the algorithm that detects the sEEG contacts (as presented at AES 2018) to improve its detection accuracy.

Funding: Please list any funding that was received in support of this abstract.: NCRR (S10RR014978) and NIH (S10RR031599, R01-NS069696, 5R01-NS060918, U01MH093765).